zBoost VMAX Mainframe Performance Enhancements

White Paper
zBoost™:
VMAX Mainframe Performance Enhancements
Abstract
This document describes zBoost™, recent Enginuity
enhancements to the VMAX product line which
improve performance and delivers full compatibility
with High Performance FICON
March 2015
Copyright © 2015 EMC Corporation. All Rights
Reserved.
EMC believes the information in this publication is
accurate as of its publication date. The information
is subject to change without notice.
The information in this publication is provided “as
is.” EMC Corporation makes no representations or
warranties of any kind with respect to the
information in this publication, and specifically
disclaims implied warranties of merchantability or
fitness for a particular purpose.
Use, copying, and distribution of any EMC software
described in this publication requires an applicable
software license.
For the most up-to-date listing of EMC product
names, see EMC Corporation Trademarks on
EMC.com.
Part Number h14023
2
Introduction
This document describes VMAX zBoostTM, a no charge microcode upgrade that improves
performance of FICON attached VMAX 40K systems up to 60% while also delivering full zHPF
support and response time reductions to both the VMAX 20K and 40K arrays. zBoost
demonstrates EMC’s commitment to protecting customers’ investments in EMC’s mainframe
storage solutions.
Users of the IBM System z platform demand high I/O throughput rates and low response times
for the mission critical applications that support their businesses. EMC has a long history of
delivering value to these environments in terms of reliability, availability, and performance. This
commitment continues with the announcement of the zBoost for VMAX.
EMC is a leading global supplier of mainframe attached storage, with a twenty year history of
supporting the world’s most demanding financial services, insurance, and government
mainframe sites. Over this period of time EMC has continually invested significant R&D
resources into mainframe storage, recently surpassing $1bn in cumulative investment at its
dedicated mainframe labs in Hopkinton, Massachusetts. The heritage of EMC Symmetrix, which
carries into the VMAX product, is that of a mainframe storage array, built like the mainframe
itself for high availability, high performance environments.
Overview
The zBoost is a no-charge Enginuity upgrade that can be applied non-disruptively to any VMAX
20K or 40K containing FICON interfaces. It delivers significant performance enhancements and
includes full zHPF compatibility 1 with support for the following additional zHPF features:
•
•
•
•
List Prefetch
Bi-Directional support
Format Writes
BSAM/QSAM/BPAM support
zBoost improves the maximum IOPS by up to 60% on the VMAX 40K, and these performance
improvements are realized when using either a single port or both ports on the two-port FICON
adapter. On the VMAX 40K zBoost reduces I/O response times by up to 40%, while on the VMAX
20K response time is reduced up to 10%. On both models response time improvements are
seen at all I/O rates across a variety of workloads.
As a result of implementing zBoost, VMAX 40K FICON configurations can support more I/O on a
given engine configuration, or require fewer FICON engines than would otherwise have been
required for a given workload. This can result in a reduction of total cost of ownership by
reducing the footprint of VMAX configurations, extending the life span of existing VMAXs, or
through realization of improved performance.
1
zHPF is a chargeable feature on VMAX, part of the Mainframe Essentials bundle
3
In addition, response time improvements will be realized during I/O bursts which previously
could have saturated the FICON adapters and resulted in queuing delays. This reduction in
queuing can improve both response time and batch run times.
zBoost performance improvements apply to both standard and zHPF FICON channels.
FICON Emulation Enhancements: VMAX 40K and 20K
zBoost contains several improvements in the design of the FICON emulation:
•
•
•
Improved pre-fetching on sequential reads
More efficient global memory access utilizing larger internal buffers
Multi-record global memory accesses for reads and writes
These changes result in lower response times throughout the entire range of I/O rates on both
the 20K and 40K VMAX models, as shown in in the performance section below.
Full compatibility with zHPF on VMAX is now delivered via zBoost. This includes:
•
List Prefetch and Bi-Directional support. These features enable a single I/O to efficiently
access discontiguous extents on a volume. This results in improved performance; for
example, in DB2 when accessing indexes with poor cluster ratios (disorganized index
scans).
•
Format Write commands. This capability improves performance of utilities, such as DB2
load, reorg, index rebuilds and restores by enabling channel programs employing format
writes to deliver large amounts of data in a single I/O.
•
Exploitation of zHPF by the BSAM, QSAM, and BPAM access methods.
VMAX 40K Performance Improvements
zBoost will benefit VMAX 40K configurations where FICON front end capacity is a bottleneck.
Simply put, zBoost adds additional processing capacity to the FICON emulation instance on the
VMAX 40K. This additional processing capacity can be exploited by a single port, up to the
saturation point of the FICON processor, or by employing both ports on the adapter card instead
of using only one port. The largest performance improvements will be realized as a result of this
change, which applies only to the VMAX 40K.
Figure 4 below illustrates that with zBoost enabled IOPS and response time benefits are seen
regardless of whether a single port or both ports are used on the FICON adapter:
4
Figure 1: zBoost equivalent IOPS and response time benefit when using one vs. two ports
Most importantly, zBoost provides both additional FICON IOPS for workload growth (or to absorb
workload peaks) and reduced response times at across all I/O rates, as figure 5 illustrates:
Figure 2: Single Engine VMAX 40K, 8 FICON paths active using zBoost
5
Finally, batch run times can be reduced since zBoost provides a higher I/O rate as well as
reduced response times for large block batch workloads as seen in figure 6 below:
Figure 3: zBoost - Large Block increased IOPS and improved response times
VMAX 20K Performance Improvements
The FICON emulation enhancements delivered by zBoost improve VMAX 20K response times by
an average of up to 10% across a variety of workloads. Figure 7 below illustrates the reduction
in response time observed on the 20K across the entire range of I/O rates in a sequential write
workload.
6
Figure 4 VMAX 20k 4 Engine response time reduction.
Implementation Considerations
Since zBoost will ‘open up’ the I/O processing capacity of the FICON adapters it is important for
existing customers employing SRDF/A to ensure the SRDF/A bandwidth is capable of supporting
additional write I/O that may be presented to SRDF adapters. Performance modeling of this
additional workload should be done and additional SRDF adapters and/or link bandwidth
should be provisioned if required prior to enabling the zBoost feature. This analysis will avoid
SRDF/A drops due to insufficient SRDF/A resources resulting from latent write demand released
as a result of performance improvements.
SRDF/S bandwidth should also be evaluated to ensure SRDF/S does not constrain the
throughput benefit of zBoost due to insufficient SRDF/S capacity.
Conclusion
In addition to the performance benefits of increased IOPS and reduced response time, the
reduction in FICON port requirements provides additional flexibility to add SRDF connectivity or
reduce engine counts in mainframe configurations. Peak workloads can now be absorbed
without the need to add engines to the configuration.
zBoost is an enhancement that is designed for easy deployment and will yield significant
performance benefits and configuration savings. EMC performance specialists have modeling
7
tools which have been updated to reflect the capabilities of zBoost. As with any configuration
change, a detailed analysis using user supplied data should be performed prior to
implementation.
8